*BSD News Article 3302


Return to BSD News archive

Path: sserve!manuel!munnari.oz.au!hp9000.csc.cuhk.hk!uakari.primate.wisc.edu!zaphod.mps.ohio-state.edu!darwin.sura.net!mips!mips!kpc!amdcad!BitBlocks.com!bvs
From: bvs@BitBlocks.COM (Bakul Shah)
Newsgroups: comp.unix.bsd
Subject: Re: System hang, working hardware guide
Message-ID: <1992Aug8.215403.2590@BitBlocks.COM>
Date: 8 Aug 92 21:54:03 GMT
References: <michaelv.713151876@test.cc.iastate.edu> <1992Aug7.160048.15061@gateway.novell.com>
Organization: Bit Blocks, Inc.
Lines: 44

terry@thisbe.npd.Novell.COM (Terry Lambert) writes:

>Anyone without the problem:

>o	What is your controller type?
>o	What is the board revision?
>o	What is the ROM revision (if any)?
>o	What are the jumper settings?
>o	What other problems do you have?


>This information will also be used to provide a "How to set up the hardware so
>386BSD will boot and run without problems the first time" guide.

My machine used to hang after heavy disk activity as well as when
I copied /sys/compile/<MACHINE>/386bsd to / right after a kernel
rebuild.  This happened on three different configurations: root &
/usr on IDE, root & /usr on SCSI, root on IDE & /usr on SCSI.
This problem went away after Bill's one liner fix.  I still don't
understand *why* that fix works but it does.  Since then the
kernel has crashed only once (when I was trying to make Type1
fonts work under X) but it did not hang even once.  And I have
rebuilt X386-1.2E-1.0.1 a couple times from scratch *while*
compiling a bunch of other things, running X etc.

My machine: a noname clone with 25Mhz 486, 128K cache, 16MB ram,
OPTi chipset, AMI bios, ET4000 SVGA card, 1024x768i 14" display,
wd(?) IDE controller & Adaptec 1542B.  Cache is enabled and Bus
clock is set at 8.33 Mhz.  ATCLK is stretch is disabled and RAM
access is 0 wait state.  GA20 line is off after power on.  All
16MBytes are cacheable.  Both Main and Video BIOS are shadowed.
I think that these are the default settings for this machine.

I did have problems unpacking things during the installation of
bin.dist -- the same hang problem.  Finally I mounted the 386bsd
filesystem on a Sun and copied the distribution by running cpio
on the Sun.  The PC did die a couple of times but thanks to NFS
statelessness & synchronous writes, the Sun cpio continued
copying once the machine came up again.  I did checksum
evertyhing just to make sure.  Until Bill's fix I unpacked
everything using this scheme.  So I'd say it is definitely worth
building a new dist.fs with just this one fix.

Bakul Shah <bvs@bitblocks.com>